Refactor Pandoc process setup using a dedicated builder class

This commit is contained in:
Thorsten Sommer 2025-05-29 16:29:47 +02:00
parent f4cac637c6
commit bb1ce9febd
Signed by: tsommer
GPG Key ID: 371BBA77A02C0108
2 changed files with 141 additions and 74 deletions

View File

@ -24,31 +24,10 @@ public static partial class Pandoc
private static readonly Version FALLBACK_VERSION = new (3, 7, 0, 2); private static readonly Version FALLBACK_VERSION = new (3, 7, 0, 2);
/// <summary> /// <summary>
/// Prepares a ProcessStartInfo for running pandoc with the given parameters. /// Prepares a Pandoc process by using the Pandoc process builder.
/// </summary> /// </summary>
/// <remarks> /// <returns>The Pandoc process builder with default settings.</returns>
/// Any local installation of pandoc will be preferred over the system-wide installation. public static PandocProcessBuilder PreparePandocProcess() => PandocProcessBuilder.Create();
/// </remarks>
/// <param name="rustService">The global rust service to access file system and data dir.</param>
/// <param name="inputFile">The input file to convert.</param>
/// <param name="outputFile">The output file to write the converted content to.</param>
/// <param name="inputFormat">The format of the input file (e.g., markdown, html, etc.).</param>
/// <param name="outputFormat">The format of the output file (e.g., pdf, docx, etc.).</param>
/// <param name="additionalArgs">Additional arguments to pass to the pandoc command (optional).</param>
/// <returns>The ProcessStartInfo object configured to run pandoc with the specified parameters.</returns>
public static async Task<PandocPreparedProcess> PreparePandocProcess(RustService rustService, string inputFile, string outputFile, string inputFormat, string outputFormat, string? additionalArgs = null)
{
var pandocExecutable = await PandocExecutablePath(rustService);
return new (new ProcessStartInfo
{
FileName = pandocExecutable.Executable,
Arguments = $"{inputFile} -f {inputFormat} -t {outputFormat} {additionalArgs ?? string.Empty} -o {outputFile}",
RedirectStandardOutput = true,
RedirectStandardError = true,
UseShellExecute = false,
CreateNoWindow = true
}, pandocExecutable.IsLocalInstallation);
}
/// <summary> /// <summary>
/// Checks if pandoc is available on the system and can be started as a process or is present in AI Studio's data dir. /// Checks if pandoc is available on the system and can be started as a process or is present in AI Studio's data dir.
@ -60,11 +39,8 @@ public static partial class Pandoc
{ {
try try
{ {
var preparedProcess = await PreparePandocProcess(rustService, string.Empty, string.Empty, string.Empty, string.Empty); var preparedProcess = await PreparePandocProcess().AddArgument("--version").BuildAsync(rustService);
var startInfo = preparedProcess.StartInfo; using var process = Process.Start(preparedProcess.StartInfo);
startInfo.Arguments = "--version";
using var process = Process.Start(startInfo);
if (process == null) if (process == null)
{ {
if (showMessages) if (showMessages)
@ -295,51 +271,7 @@ public static partial class Pandoc
} }
} }
/// <summary> public static async Task<string> GetPandocDataFolder(RustService rustService) => Path.Join(await rustService.GetDataDirectory(), "pandoc");
/// Reads the os platform to determine the used executable name.
/// </summary>
private static string PandocExecutableName => CPU_ARCHITECTURE is RID.WIN_ARM64 or RID.WIN_X64 ? "pandoc.exe" : "pandoc";
/// <summary>
/// Returns the path to the pandoc executable.
/// </summary>
/// <remarks>
/// Any local installation of pandoc will be preferred over the system-wide installation.
/// When a local installation is found, its absolute path will be returned. In case no local
/// installation is found, the name of the pandoc executable will be returned.
/// </remarks>
/// <param name="rustService">Global rust service to access file system and data dir.</param>
/// <returns>Path to the pandoc executable.</returns>
private static async Task<PandocExecutable> PandocExecutablePath(RustService rustService)
{
//
// First, we try to find the pandoc executable in the data directory.
// Any local installation should be preferred over the system-wide installation.
//
var localInstallationRootDirectory = await GetPandocDataFolder(rustService);
try
{
var executableName = PandocExecutableName;
var subdirectories = Directory.GetDirectories(localInstallationRootDirectory);
foreach (var subdirectory in subdirectories)
{
var pandocPath = Path.Combine(subdirectory, executableName);
if (File.Exists(pandocPath))
return new(pandocPath, true);
}
}
catch
{
// ignored
}
//
// When no local installation was found, we assume that the pandoc executable is in the system PATH.
//
return new(PandocExecutableName, false);
}
private static async Task<string> GetPandocDataFolder(RustService rustService) => Path.Join(await rustService.GetDataDirectory(), "pandoc");
[GeneratedRegex(@"pandoc(?:\.exe)?\s*([0-9]+\.[0-9]+(?:\.[0-9]+)?(?:\.[0-9]+)?)")] [GeneratedRegex(@"pandoc(?:\.exe)?\s*([0-9]+\.[0-9]+(?:\.[0-9]+)?(?:\.[0-9]+)?)")]
private static partial Regex PandocCmdRegex(); private static partial Regex PandocCmdRegex();

View File

@ -0,0 +1,135 @@
using System.Diagnostics;
using System.Reflection;
using System.Text;
using AIStudio.Tools.Metadata;
using AIStudio.Tools.Services;
using SharedTools;
namespace AIStudio.Tools;
public sealed class PandocProcessBuilder
{
private static readonly Assembly ASSEMBLY = Assembly.GetExecutingAssembly();
private static readonly MetaDataArchitectureAttribute META_DATA_ARCH = ASSEMBLY.GetCustomAttribute<MetaDataArchitectureAttribute>()!;
private static readonly RID CPU_ARCHITECTURE = META_DATA_ARCH.Architecture.ToRID();
private string? providedInputFile;
private string? providedOutputFile;
private string? providedInputFormat;
private string? providedOutputFormat;
private readonly List<string> additionalArguments = new();
private PandocProcessBuilder()
{
}
public static PandocProcessBuilder Create() => new();
public PandocProcessBuilder WithInputFile(string inputFile)
{
this.providedInputFile = inputFile;
return this;
}
public PandocProcessBuilder WithOutputFile(string outputFile)
{
this.providedOutputFile = outputFile;
return this;
}
public PandocProcessBuilder WithInputFormat(string inputFormat)
{
this.providedInputFormat = inputFormat;
return this;
}
public PandocProcessBuilder WithOutputFormat(string outputFormat)
{
this.providedOutputFormat = outputFormat;
return this;
}
public PandocProcessBuilder AddArgument(string argument)
{
this.additionalArguments.Add(argument);
return this;
}
public async Task<PandocPreparedProcess> BuildAsync(RustService rustService)
{
var sbArguments = new StringBuilder();
if(!string.IsNullOrWhiteSpace(this.providedInputFile))
sbArguments.Append(this.providedInputFile);
if(!string.IsNullOrWhiteSpace(this.providedInputFormat))
sbArguments.Append($" -f {this.providedInputFormat}");
if(!string.IsNullOrWhiteSpace(this.providedOutputFormat))
sbArguments.Append($" -t {this.providedOutputFormat}");
foreach (var additionalArgument in this.additionalArguments)
sbArguments.Append($" {additionalArgument}");
if(!string.IsNullOrWhiteSpace(this.providedOutputFile))
sbArguments.Append($" -o {this.providedOutputFile}");
var pandocExecutable = await PandocExecutablePath(rustService);
return new (new ProcessStartInfo
{
FileName = pandocExecutable.Executable,
Arguments = sbArguments.ToString(),
RedirectStandardOutput = true,
RedirectStandardError = true,
UseShellExecute = false,
CreateNoWindow = true
}, pandocExecutable.IsLocalInstallation);
}
/// <summary>
/// Returns the path to the pandoc executable.
/// </summary>
/// <remarks>
/// Any local installation of pandoc will be preferred over the system-wide installation.
/// When a local installation is found, its absolute path will be returned. In case no local
/// installation is found, the name of the pandoc executable will be returned.
/// </remarks>
/// <param name="rustService">Global rust service to access file system and data dir.</param>
/// <returns>Path to the pandoc executable.</returns>
private static async Task<PandocExecutable> PandocExecutablePath(RustService rustService)
{
//
// First, we try to find the pandoc executable in the data directory.
// Any local installation should be preferred over the system-wide installation.
//
var localInstallationRootDirectory = await Pandoc.GetPandocDataFolder(rustService);
try
{
var executableName = PandocExecutableName;
var subdirectories = Directory.GetDirectories(localInstallationRootDirectory);
foreach (var subdirectory in subdirectories)
{
var pandocPath = Path.Combine(subdirectory, executableName);
if (File.Exists(pandocPath))
return new(pandocPath, true);
}
}
catch
{
// ignored
}
//
// When no local installation was found, we assume that the pandoc executable is in the system PATH.
//
return new(PandocExecutableName, false);
}
/// <summary>
/// Reads the os platform to determine the used executable name.
/// </summary>
private static string PandocExecutableName => CPU_ARCHITECTURE is RID.WIN_ARM64 or RID.WIN_X64 ? "pandoc.exe" : "pandoc";
}